The analysis of sound symbolism in poetry is one of the more promising
applications of computational methods. This paper proposes using data
base software with spreadsheet capabilities to give maximum versatilit
y in the examination of consonant alliteration. In this case the datab
ase is drawn from a 10th century anthology of classical Japanese verse
called the Kokinshu. In recent years scholars have pointed out a few
obvious examples of sound symbolism in Kokinshu poetry. This study att
empts to show that with these few notable exceptions, poetry of the pe
riod seems to have striven toward a balance in sound, avoiding techniq
ues such as word initial alliteration which might call attention to it
self.